Statistical methods for high-throughput experimental data

Description

Invited talk held at the Session "Statistics for a wise use of machine learning"
Abstract: Exploiting reactions from high-throughput experimentation using machine learning techniques is becoming state-of-the-art in organic chemistry. Many valuable data sets are being generated in order to learn about reaction conditions that are crucial for the outcomes of chemical reactions such as yields
or selectivities. However, it is typically ignored, that the data from designed experiments inherit a very specific structure that needs to be taken into account in the analysis in order to make appropriate conclusions. On the example of the data from Buchwald-Hartwig Amination, which were collected for nearly 4000 reaction conditions, we demonstrate the shortcomings of used machine learning techniques and suggest a statistically rigorous approach to the analysis of such data from high-throughput experiments.

Period	29 Jun 2024
Event title	International Symposium on Nonparametric Statistics 2024
Event type	Conference
Location	Braga, PortugalShow on map
Degree of Recognition	International